Two methods have traditionally been used to produce complex polymers such as polypeptides. One method relies on relatively uncontrolled polymerization reactions, wherein monomer subunits react to produce large polymers. Using this method, polydisperse macromolecules such as polypeptides and plastics (e.g., polyethylene or nylon) can be produced from monomeric residues such as amino acids or small aliphatic or aromatic organic molecules, respectively. While such polydisperse macromolecules can be relatively easy to produce, the polymeric macroscopic products are not homogeneous at the molecular level but are mixtures of polymers of different lengths and even different composition, e.g., in a random copolymer. Furthermore, the similarity of the homologs produced in such polydisperse preparations makes it difficult or impossible to obtain a single high molecular weight product in pure form.
The second method utilized to obtain complex macromolecules has been the sequential assembly of reversibly protected monomers. This approach can be used to obtain products of defined, typically linear, structure. Unfortunately, the method is limited in the size and, most critically, the complexity of molecules that can be produced. For example, synthesis of defined polypeptides or proteins larger than about 50-80 amino acid residues has been beyond the reach of this technology. Condensation of pre-purified protected peptides, two at a time, is limited by the insolubility of large protected fragments. As a result, synthesis of homogeneous, linear polypeptides, for example, is limited to an upper limit of about 100 amino acid residues.
Mutter et al. (Proteins:Structure, Function and Genetics (1989) 5:13-21) have synthesized branched chain polypeptides by step-wise coupling of protected amino acids to a synthetic, protected, resin-bound peptide template during solid-phase peptide synthesis. Deprotection and cleavage was required to obtain a soluble template-assembled synthetic protein. Also using step-wise, solid phase peptide synthesis, Tam and Zavala (J. Immunol. Meth. (1989) 124:5261) have built branched chain "lysine tree" templates with peptide branches, referred to as multiple antigen peptides, which were subsequently obtained in soluble, crude form after HF deprotection and cleavage.
Since protecting groups used in polypeptide synthesis generally decrease solubility of the protectable molecule, the ability to condense unprotected polypeptides would provide an improvement in the solubility problem encountered using protected precursors, as well as minimize harsh deprotection methods needed to achieve a final product. However, the use of unprotected precursors raises the seemingly insurmountable problem of regiospecificity. Therefore, attempts have been made to use regiospecific condensation of unprotected fragments through the use of chemoselective ligation.
Chemoselective ligation requires the use of complementary pairs of reactive groups present at specific sites on the precursor molecules that are being joined. The use of reactive groups having complementary chemical reactivity, such as a thiol group and a bromoacetyl group, results in the formation of a bond in a regiospecific manner. For example, thiol-type chemoselective ligations have been used to prepare multi-antigenic peptides. In an attempt to avoid harsh deprotection methods and formation of impure products caused by possible steric hindrance between closely spaced growing peptide arms during step-wise solid phase synthesis, Drijfhout and Bloemhoff (Int. J. Peptide Protein Res. (1991) 37:27-32) used thiol-type chemoselective coupling by synthesizing a branched "octa-amino lysine tree" peptide, whose deprotected amino groups were extended to contain protected sulfhydryl groups (S-acetylmercaptoacetyl) for subsequent coupling to an appropriately modified sulfhydryl-containing antigenic peptide. However, the product obtained had poor characteristics as defined by high performance liquid chromatography and was not fully characterized. More recently, thiol chemistry was used to prepare, in a two-fragment condensation, a totally synthetic, linear, functional HIV protease analog (Schnolzer and Kent (1992) Science 256:221-225).
Unfortunately, thiol chemistry is not completely specific. It is well known, for example, that thiol groups can participate in disulfide bond shuffling. In addition, alkylating agents such as bromoacetyl and maleoyl can react with nucleophilic amino acid groups other than a thiol group. For example, bromoacetyl can react with the thioether side chain of methionine residues, thus limiting the homogeneity and/or complexity of design of the desired product.
Thus, a need exists for preparations, particularly homogeneous preparations, of easily synthesized macromolecules of defined structure having stable ligation linkages and for reagents and methods for constructing these preparations of macromolecules that provide ease, rapidity and mildness of synthesis; essentially quantitative yields; versatility in design; and applicability to construction using a diversity of biochemical classes of compounds. In addition a need exists for macromolecules, particularly homogeneous macromolecules, of defined structure that can be designed for desired activity, solubility, conformation and other desirable properties; that present components, such as peptides or oligonucleotides, in non-linear, polyvalent form; that provide higher binding affinity and specificity of interaction; and, in the case of homogeneous macromolecules, can made available as homogeneous preparations. Furthermore the need exists for libraries of macromolecules such as peptides or oligonucleotides having advantages discussed herein, The present invention satisfies these needs and provides related advantages as well.